Progressive Collapse Analysis of Steel Truss Bridges

Transcription

1 Proc. Schl. Eng. Tokai Univ., Ser. E 35 (2010) (2010) Progressive Collapse Analysis of Steel Truss Bridges by Akihiro MANDA *1 and Shunichi NAKAMURA *2 (Received on Sep. 27, 2010 and accepted on Nov. 17, 2010) Abstract Deteriorated steel truss bridges have caused catastrophic problems in the USA and Japan. Progressive collapse analysis is carried out for the three continuous steel truss bridges using large deformation elastic plastic analysis. It is intended to clarify how the live load intensity and distribution affect structural safety and ductility for these two truss bridges. Although the collapse process is different depending on live load distribution and length of the spans, the steel truss bridge collapses due to plastic buckling or elastic buckling. It is found that ductility of Model Bridge-B with a span ratio of 1:1.3:1 is larger than that of Model Bridge-A with a span ration of 1:2:1. Keywords: Progressive collapse, Large deformation elastic plastic analysis, Ductility, Structural safety 1. Introduction The steel truss bridge I-35W over the Mississippi River in Minneapolis, Minnesota, United States, suddenly collapsed on the 1st of August 2007 and there were many victims by this disaster 1). The I-35W had 3 continuous spans, the side-span of 81m and the main-span of 139m. The span ratio was 1:1.7:1. The report about collapse of the I-35W clarifies that dead load increased by several times of slab maintenances. The thickness of gusset plate was half of the design. Corrosion of the gusset plate was one of the possible casuse of the collapse. In addition, on the day the I-35W collapsed, there were construction materials and heavy machinery on the truss bridge for the maintenance. Construction materials were 174t weigh and heavy machinery was 89t weigh, total weight of 263t was distributed near the intermediate support. These factors are the possible causes of the collapse of the I-35W 2). In Japan, there are many aging truss bridges as well and they need prompt inspection, reinforcement and maintenance. On the Kiso River Bridge and the Honjo Bridge the diagonal member penetrates the RC slab and it fractured because of the corrosion 3). Nagatani et al. studied about redundancy of truss bridges 4). Kasano et al. focused on the gusset plate. They clarify that, when the gusset plate have less thickness and *1 Graduate Student, Course of Civil Engineering *2 Professor, Department of Civil Engineering initial displacement, it causes increasing of strain 5). However, there is no research about how the live load intensity and distributions affect the plasticity, the final deformation and the collapse load. Moreover, there is no research about the affect of the span ratio. In this research it is intended to clarify these factors for three continuous span truss bridges. It is highly difficult to find the collapse loads and mechanism because non-linearity of materials. Geometry must be considered in complicated structures. Therefore, structural models are simplified as two dimensional and gusset plates are not considered. The progressive collapse analysis is conducted for the four load cases and for the two bridge models with different span ratios by large deformation elastic-plastic analysis. Then the collapse process, the collapse load and the final deformation are obtained. Furthermore, it is studied how the span ratio and the live load distribution affect the truss bridge ductility. 2. Structural Model 2.1 Structural model Fig.1 and Fig.2 show the bridge model, a Warren truss bridge with three continuous spans with a total length of 230m. Model Bridge-A has a span ratio of 1:2:1 with a main-span of 115m, a side span of 57.5m and a height of 10m. Although Model Bridge-A is nearly the same as the I-35W, the objective of this study is not to investigate the collapse causes of the I-35W but to conduct the progressive collapse Vol.,

2 Progressive Collapse Analysis of Steel Truss Bridges Akihiro MANDA and Shunichi NAKAMURA 10,000 A F 57,500 B G E 11,500=115, ,000 mm C 57,500 D Fig.1 Side View Model Bridge-A (1:2:1) 10,000 A F H B G E 69,000 11,500=92, ,000 mm Fig.2 Side View Model Bridge-B (1:1.3:1) C 69,000 D 510 Stress σ y ,850 9,000 2,850 2, ,00 Strain 1, ,000 Fig.3: Cross section (mm) σ y Fig.4: Stress-Strain curve of steel Table1 Yield Stress, Yield Strain and Young s Modulus SM490Y SS400 tensile strength(n/mm 2 ) σ u=490 σ u=400 yield stress (N/mm 2 ) σ y =355 σ y =235 yield strain (μ) ε y =1775 ε y=1175 Young's modulus (N/mm 2 ) Es= Es= E2= E2= Upper chord Lower chord Diagonal Fig.5: Cross section of structural members (mm) 477 P D P L1 P L2 P L3 10m Case-1 P D +P L1 Case-2 P D +P L2 Case-3 P D +P L3 P L4 Fig.6 Load cases Case-4 P D +P L4 28 Proceedings of the School of Engineering, Tokai University, Series E

3 Akihiro MANDA and Shunichi NAKAMURA Progressive Collapse Analysis of Steel Truss Bridges A U-1 U-2 U-3 U-1 U-2 D-4 D-5 D-4 D-5 L-6 L-7 L-8 A L-6 L-7 F B G E F H B G U-3 L-8 E Fig.7 Typical members Model Bridge-A (1:2:1) Fig.8 Typical members Model Bridge-B (1:1.3:1) Table 2 Results of the allowable stress design method Bridge-A Table 3 Results of the allowable stress design method Bridge-B Size (mm) Steel Member σ/σa UFlg Web LFlg Grade No. P D +P L1 P D +P L2 P D +P L3 P D +P L4 600*26 530*26 480*33 SS400 U *22 380*22 320*24 SM490Y U *28 510*28 480*33 SM490Y U *20 450*15 477*20 SM490Y D *25 460*15 477*25 SM490Y D *22 480*22 600*18 SS400 L *22 480*22 600*18 SM490Y L *22 500*22 620*20 SS400 L analysis for different live load distribution and span ratio. Model Bridge-B has a span ratio of 1:1.3:1 with a main-span of 92m, a side of 69m and a height of 10m. The length of each upper chord and lower chord is 11.5m. Fig.3 shows the cross section of structural members and deck system. In this study RC slab is assumed. The steels are either SM490Y or SS400 and the stress-strain curve is shown in Fig.4 6). Table 1 shows the yield stress, yield strain and Young s modulus of SM490Y and SS400. The ultimate strain is set at 5%. Fig.5 shows typical cross sections of the structural members. To simplify this truss bridge, all diagonals are assumed to have rectangular cross sections. The sizes and steel grades of the members were determined by allowable stress design method based on the Japanese current design specification 6). 2.2 Preliminary design of truss members Static structural analysis was conducted for the dead load (P D ) and the design live loads (P L : combination of uniformly-distributed of 3.5 kn/m 2 and concentrated loads of 10.0 kn/m 2 with a longitudinal width of 10m). Then sectional forces and deformations were obtained. The design live load of B-type of the Japanese specifications for road bridges is adopted in this study. Four load cases, P D +P L1, P D +P L2, P D +P L3 and P D +P L4 are considered in this progressive collapse analysis. As shown in Fig.6, P L1 is the live lodes distributed in full spans, P L2 is that distributed only in the main span, P L3 is that distributed only in the side span, and P L4 is that distributed only near the intermediate support B. Weight per unit length for the dead load is 60.56kN/m. The live load intensity is a combination of kn/m (uniformly distributed loads), and kn/m (concentrated loads with length of 10m). Thickness and steel grade of the truss members were then determined by the allowable stress method for the two bridge models. Fig.7 and Fig.8 show the location of the designed members. Table 2 and Table 3 show design stresses Size (mm) Steel Member σ/σa UFlg Web LFlg Grade No. P D +P L1 P D +P L2 P D +P L3 P D +P L4 420*20 360*20 300*24 SS400 U *22 380*22 320*24 SM490Y U *28 510*28 480*33 SM490Y U *14 430*15 450*14 SS400 D *20 450*15 477*20 SM490Y D *22 480*22 600*18 SS400 L *22 480*22 600*18 SM490Y L *22 480*22 600*18 SM490Y L over the allowable stresses in four load cases. These tables show that almost all the members stresses are over 80% of allowable stress. 3. Result of progressive collapses analysis Progressive collapse analysis is conducted in this section. Member s section is divided as fiber elements; the flange is divided into five layers in depth and five slices in width. The web is divided into fifteen layers in depth and fifteen slices in width. Each fiber follows the stress vs. strain relation of Fig.4. The analytical procedure is as follows. First the dead load is applied, and then the live load is increased incrementally until the bridge collapses. This can be expressed as P D + k P L, where k is the load amplification coefficient. Analysis includes elastic-plastic properties of steel and large deformation effect. Calculation was carried out by FEM program, FORUM8 FRAME (3D). 3.1 Analysis results of Model Bridge-A Collapse processes and final deformations of the four load cases are shown in Fig.9. In Case-1, the members near support B became yield in tension at first. Then, tensile yield expanded at the mid-span. Next, compression yield appeared near intermediate support and in span center. Finally, when the load amplification coefficient k became 5.45, the upper chord buckled at the mid-span. Maximum strain of tensile member was 4.3% at the upper chord near the support B. In Case-2, tensile yield appeared in the side span upper chord at first. Then, tensile yield expanded at the mid-span. When the load amplification coefficient k became 5.45, the upper chord suddenly buckled at the mid-span. Maximum strain of tensile member was 3.1% at upper chord in the side span. In Case-3, the tensile diagonal yielded first near the intermediate support B. However, tensile yield didn t expand Vol.,

4 Progressive Collapse Analysis of Steel Truss Bridges Akihiro MANDA and Shunichi NAKAMURA P D P L1 Case 1 Tensile yield Compression yield Compressive Buckling P D P L1 P D P L1 P D P L3 Case 3 P D P L1 P D P L3 P D P L1 P D P L3 P D P L1 P D P L3 P D P L1 P D P L3 P D P L1 P D P L1 P D P L4 Case 4 P D P L2 Case 2 P D P L4 P D P L2 P D P L4 P D P L2 P D P L4 P D P L2 P D P L4 P D P L2 P D P L4 Fig.9 Collapse process and final deformation (Model Bridge-A) 30 Proceedings of the School of Engineering, Tokai University, Series E

5 Akihiro MANDA and Shunichi NAKAMURA Progressive Collapse Analysis of Steel Truss Bridges Case1 Case2 Case3 Case ε k δ/δy Case1 at E Case2 at E Case3 at F Case4 at G Fig.10 Stress-Strain (Compression) σ/σy Fig.11 Load amplification vs. displacement σ/σy 0.50 Case1 Case2 Case3 Case ε Fig.12 Stress-Strain (Tension) until the bridge collapsed. When the load amplification coefficient k became 5.90, compressive members started yield. With a small load increment k of 0.1, the upper chord buckled. Lower chord didn t yield. Maximum strain of tensile members was 2.3% at diagonal member the near the intermediate support B. In Case-4, the diagonal member near the support B became yield at first. Then, compressive yield appeared near the support B, and tensile yield also expanded. When the load amplification coefficient k became 9.43, the diagonal member near the support B buckled. However, the displacement wasn t large when the diagonal member buckled near support B. And then, when load amplification coefficient k is 9.44 with a small load increment after buckling, the main span fell downwards. Maximum strain of tensile member was 4.0% at the diagonal member near the support B. Fig.10 shows the compressive stress-strain curve of the buckled members. The bridge collapsed at σ/ σ y of in Case-1, in Case-3 and in Case-4. These results show that the model truss bridge collapsed because of plastic buckling in these cases. In Case-2 the bridge collapsed at σ/ σ y of -0.92, which means the bridge collapsed because of elastic buckling Fig.11 shows relationships of load and displacement at the critical points. The load amplification coefficient k instead of load itself is used so as to compare each case s relationship of load-displacement. Displacements are shown as non-dimensional parameter. In all the cases, vertical displacement sharply increased after the buckling. In Case-4, the yield point appeared near the node G at the coefficient k of 6.0, and then δ/δ y started to increase until coefficient k of Fig.12 shows the stress and strain relation of the critical tensile members. To compare the steel members with different yield strength, these figures vertical axes are displayed as non-dimensional parameter. Fig.12 shows that the maximum strain didn t reach the design ultimate strain of 5%, and therefore the collapse was not caused by tensile failure. 3.2 Analysis results of Model Bridge-B Fig.13 shows the results of progressive collapse analysis and the final deformation of Model Bridge-B for the four load cases. In Case-1, the diagonal member near intermediate support B became yield in tension at first. The first compressive yield member was the lower chord near the support B. Both tensile and compressive yield members expanded at the mid-span and near the support B. Finally, when the load amplification coefficient k became 7.15, the diagonal member near support B buckled. The maximum strain of tensile member was 4.0% at the diagonal member near the support B. In Case-2, the lower chord at the mid-span became yield in tension when coefficient k of 5.0. When the coefficient k became 7.0, the compressive yield members appeared at the mid-span and near intermediate support B. Finally, when the load amplification coefficient k became 7.5, the upper chord at the mid-span buckled. The maximum strain of tensile member was 3.9% at the lower chord in the center span. In Case-3, the diagonal member near the support B became yield in tension at first. Next, compressive yield appeared in the side span. Then, tensile yield and compressive yield expanded in the side span. When coefficient k became 6.5, the strain of upper chord in the side span sharply increased and buckled. In this case, maximum strain of tensile member was 5.0% at the diagonal member near the support B. It was thought that this maximum strain of tensile member occurred after the buckling. In Case-4, the diagonal member near intermediate support B became yield in tension when load amplification Vol.,

6 Progressive Collapse Analysis of Steel Truss Bridges Akihiro MANDA and Shunichi NAKAMURA Case 1 Tensile yield Compression yield Compressive Buckling Case 3 Case 2 Case 4 Fig.13 Collapse process and final deformation (Model Bridge-B) 32 Proceedings of the School of Engineering, Tokai University, Series E

7 Akihiro MANDA and Shunichi NAKAMURA Progressive Collapse Analysis of Steel Truss Bridges k Case1 Case2 Case3 Case ε Fig.14 Stress-Strain (Compression) Fig.15 Load amplification vs. displacement σ/σy δ/δy Case1 at H Case2 at E Case3 at F Case4 at G 0.50 Case1 Case2 Case3 Case ε Fig.16 Stress-Strain (Tension) coefficient k reached 7.0. Compressive yield appeared when coefficient k was 9.0 near the support B. Finally, when the load amplification coefficient k became 11.0, the diagonal member near the support B buckled. In this case only ninemembers became yield when the live load was applied incrementally. After the buckling of the diagonal near the support B, when the increment k became 0.1, the center span fell underneath. This deformation is similar to the collapse of the I-35W in Minnesota, United States. Maximum strain of tensile member was 4.7% at the diagonal member near the support B. The stress-strain relationships of the buckled members are shown in Fig.14. Model Bridge-B collapsed at σ/ σ y of in Case-1 and Case-2, in Case-3 and in Case-4. These results show that collapse occurred by plastic buckling. σ/σ y The difference of collapse processes between Model Bridge-A and Model Bridge-B are as follows. In Case-1, Model Bridge-A collapsed because of the buckling of upper chord in mid-span, whereas Model Bridge-B collapsed because of the buckling of the diagonal member near the support B. In Case-2, Model Bridge-A collapsed due to elastic buckling. Model Bridge-B collapsed due to plastic buckling. In Case-3, Model Bridge-A has no yielded lower chords in side span, however Model Bridge-B has some yielded lower chords in side span. Fig.15 is the relationships of load and displacement at the critical points. Coordinate axes are the same as Model Bridge-A. In Case-1, δ/δ y was in Case-2 δ/δ y was when the bridge collapsed. In Model Bridge-A, after the yield load, δ/δ y sharply increased in all four cases. However δ/δ y increased gradually in Model Bridge-B. These results show that collapse processes of Model Bridge-B are more gradual than Model Bridge-A. Fig.16 is the stress-strain relationship of tensile members, which yielded first in each case. These members have the maximum tensile strain in each case. The maximum strain didn t reach the design ultimate strain of 5% in Case-1 (4.0%), Case-2 (3.9%) and Case-4 (4.7%). In Case-3, the maximum strain was 5.0%. The ultimate strain of steel in general is more than 5.0%. However, there may be some scatter or residual stresses. So, in this study the ultimate strain was set at 5.0% as a safer value. 4. Ductility evaluation As previously mentioned, progressive collapse analysis was conducted for the two bridge models with different span ratios. Processes, section forces and displacement are obtained by progressive collapse analysis. Depending on the span ratio and live load distributions, collapse processes are different in each case. In this chapter, ductility evaluation is conducted for each bridge models for four load cases. Ductility of the Warren truss bridge in each model and each live load case were evaluated as bellow. The live load amplification coefficient is expressed as ky when a first yield member appeared. The live load amplification coefficient is expressed as ku when buckling occurred. Ductility μ is defined as the fraction of ky and ku. Table 4 and Table 5 are the ductility evaluation. In Case-1, the difference of ductility μ between Model Bridge-A and Model Bridge-B is In Case-2, difference of μ is 0.14 and in Case-3 it is 0.3. In Case-4 ductility μ are same. However the value of ky and ku in Model Bridge-B are larger than Model Bridge-A. From these results, it is found that Vol.,

8 Progressive Collapse Analysis of Steel Truss Bridges Akihiro MANDA and Shunichi NAKAMURA Table 4 Ductility evaluations Model Bridge-A Model-A ky ku μ=ku/ky Case Case Case Case Table 5 Ductility evaluations Model Bridge-B Model-B ky ku μ=ku/ky Case Case Case Case Bridge Name / Load Case No. Model A Model B Table 6 Number of yield members Case1 Case2 Case3 Case Model Bridge-B is more ductile than Model Bridge-A. Table 6 shows the number of yield members when the buckling occurred. For Case-1 in Model Bridge-A and Case-2 in both models, bridge collapsed due to the buckling of upper chord at mid-span. In these cases, number of yield members was more than 16. The number of yield members was 13 in Case-1, Model Bridge-B. In Case-4, both Model Bridge-A and Model Bridge-B collapsed due to the buckling of diagonal near support B. Number of yield members was 6 in Model Bridge-A and 9 in Model Bridge-B. These results show that, when the live load was distributed in full span or in main span (Case-1 and Case-2), many members reached to the yield, and redistribution of loads can be seen. On the other hand, when the live load was distributed only near the intermediate support (Case-4), fewer members yielded and redistribution of load is small. 5. Conclusions In this study,the collapse process is clarified by the large deformation elastic plastic method. The collapse process is different depending on live load distribution and length of the each span. Especially, we tried to clarify the collapse process, the collapse load and the final deformation. Furthermore, it is intended to clarify how the span ratio and the live load distribution affect the truss bridge ductility. The final load amplification coefficient k is over 5.4 in all cases. This shows that the designed bridge has ample safety against the live load. For Model Bridge-A the span ratio of 1:2:1, the bridge collapse due to the compressive elastic or plastic buckling in all four load cases. Maximum strain of tensile member was less than 5% in the four cases. It clarify that the bridge didn't collapse due to the tensile failure. Model Bridge-B with the span ratio of 1:1.3:1 collapses due to the compressive plastic buckling in the four load cases. Maximum strain of tensile member was less than 5% in the four cases. The bridge didn't collapse due to the tensile failure. Displacement increased more gradually in Model Bridge-B. Ductility μ is defined as a fraction of ky and ku. As the result of ductility, it is found that Model Bridge-B is more ductile than Model Bridge-A. As the span ratio of Model Bridge-B is more commonly used, it is not difficult to understand that Model Bridge-B is more ductile. Considering the number of yield members when the buckling occurred, when the live load is distributed in full span or in main span, many members yielded and redistribution of load is formed. Also redistribution of load is larger in Model Bridge-B. On the other hand, when the live load distributed only near the intermediate support, less members yield and redistribution of load is small. It is found that common span ratio of Model Bridge-B is more ductile than Model Bridge-A. And has higher redistribution of loads. Other researchers have concerned about redundancy analysis or analysis of the gusset plate for the truss bridge collapse problem. On the other hand, it is found in our studies that collapse process and buckling member significantly depend on the live load distribution and the span ratio. Reference 1) National Transportation Safety Board: High Way Accident Report, Collapse of I-35W Highway Bridge Minneapolis, Minnesota August 1, 2007, ( ) NTSB/HAR-08/03PB , 2) H.Hatano, S.Murakami, Y.Rokugou, T.Yoda: Summary of NTSB Report on I-35W Bridge in Minneapolis, Bridge and Foundation, (2010.7), pp ) T.Tamakoshi: Results of the technical investigation group for the bridge collapse in USA, Constructional Management Technology, (2008.1),pp ) H.Nagatani, N.Akashi, T.Matsuda, M.Yasuda, H.Ishii, M.Miyamori, Y.Obata, H.Hirayama, Y.Okui: Redundancy analysis of Japanese steel truss bridges, Proceedings of Japan Civil Engineering Association-A, Vol.65 No.2, (2009.5), ) H.Kasano, T.Yoda: Collapse mechanism and evaluation of node damages of I-35W Bridge, Mineapolis, USA, Proceedings of Japan Civil Engineering Association-A, Vol.66, No.2, (2010), ) Japan Road Association:Specifications for Road Bridges, ⅠGeneral,ⅡSteel Bridges, (2002.3) 34 Proceedings of the School of Engineering, Tokai University, Series E